Comment by bondolo

2 months ago

I had a conversation the other day at a birthday party with my friend's neighbour from the building. The fellow is a semi-retired (FIRE) single guy. We started with a basic conversation but then he started talking about what he interested in and it became almost unintelligible. I kept having to ask him to explain what he was talking about but was increasingly unsuccessful as he continued. Sure enough though, he described that he spent significant time talking with "AIs" as he called them. He spends many hours a day chatting with ChatGPT, Grok and Gemini (and I think at least one other LLM). I couldn't help thinking "Dude, you have fucked up your brain." His insular behaviour and the feedback loop he has been getting from excessive interaction with LLMs has isolated him and I can't help but think that will only get worse for him. I am glad he was at the party and getting some interaction with humans. I expect that this type of "hikikomori" isolation will become even more common as LLMs continue to improve and become more pervasive. We are are likely to see this become a significant social problem in the next decade.

What was the nature of his interests, if you don't mind sharing? I'm always curious about how these things develop -- makes it easier to recognize.

Seems like a lot of them fall into either "I'm onto a breakthrough that will change the world" (sometimes shading into delusion/conspiracy territory), or else vague platitudes about oneness and the true nature of reality. The former feels like crankery, but I wonder if the latter wouldn't benefit from some meditation.

  • It was a mix of mystical philosophy and transhumanism and he does think that "the world is on the edge of a breakthrough" but he sees it as emergent. It is not something he is personal creating just something he believes is imminent and he is one of the first people to recognise it.

    • Thanks -- so a little of column A, a little of column B? Kind of feels similar to how early societies built whole religious practices out of interpreting stochastic phenomena like knucklebones or entrails, only supercharged because this particular viscera seems to talk back!

    • Oh weird, was he also bad at socializing?

      I often have to remind myself of the quote "Talk to a man about himself and he will listen for hours" when socializing to remember to ask questions and let the other party explore whatever topic/situation they are into. It seems like AI conversations are so one-sided a person might forget to cede the floor entirely.

Did he refer to the AI with a name? How much of a relationship did he have with his? I have multiple friends that have named their ChatGPT, and they refer to it in conversation, like "oh yeah, Sarah told me this or that the other day", except Sarah (names changed) is an LLM.

I'm worried about our future.

...except I went over to ChatGPT and asked it to project what the future looks like in seven years rather than think about it myself. Humanity is screwed.